An unsupervised competitive neural network algorithm for clustering mixtures of Gaussian probability density functions is proposed. The algorithm based on centroid neural network with Bhattacharyya distance is evaluated in the context of speech recognition and the results show that it can reduce the Gaussian mixtures by almost 60% over the k-means algorithm.
References
-
-
1)
-
E.W. Forgy
.
Cluster analysis of multivariate data: efficiency vs interpretability of classifications.
Biometrics
-
2)
-
D.-C. Park ,
Y.-J. Woo
.
Weighted centroid neural network for edge reserving image compression.
IEEE Trans. Neural Netw.
,
1134 -
1146
-
3)
-
Watanabe, T., Shinoda, K., Takagi, K., Iso, K.I.: `High speed speech recognition using tree structured probability density function', ICASSP-95, 1995, Detroit, MI, USA, 1, p. 556–559.
-
4)
-
A. Kannan ,
M. Ostendorf ,
J.R. Rohlicek
.
Maximum likelihood clustering of Gaussians for speech recognition.
IEEE Trans Speech Audio Process.
,
3 ,
453 -
455
-
5)
-
K.W.B. Mak ,
E. Barnard
.
(1998)
Phone clustering using the Bhattacharyya distance.
-
6)
-
D.-C. Park
.
Centroid neural network for unsupervised competitive learning.
IEEE Trans. Neural Netw.
,
2 ,
520 -
528
-
7)
-
K.W.B. Mak ,
E. Bocchieri
.
Subspace distribution clustering hidden Markov model.
IEEE Trans. Speech Audio Process.
,
3 ,
264 -
275
http://iet.metastore.ingenta.com/content/journals/10.1049/el_20030247
Related content
content/journals/10.1049/el_20030247
pub_keyword,iet_inspecKeyword,pub_concept
6
6